Soil Science Society of America Journal

Soil Sci. Soc. Am. J. 82:939–948 doi:10.2136/sssaj2018.03.0098 Received 9 Mar. 2018. Accepted 23 May 2018. *Corresponding author (tunsisa.hurisso@gmail.com). © Soil Science Society of America, 5585 Guilford Rd., Madison WI 53711 USA. All Rights reserved.

Repeatability and Spatiotemporal Variability of Emerging Soil Health Indicators Relative to Routine Soil Nutrient Tests

Soil & Water Management & Conservation

Ideal indicators suitable for routine soil health evaluation would be rapid, cost-effective, sensitive to management, and exhibit low analytical variabil-ity. Permanganate-oxidizable c (POXc), mineralizable c, and soil protein are rapid and emerging soil health indicators of labile organic matter (OM), but their repeatability and spatiotemporal variability remain largely unknown. We examined the repeatability and spatiotemporal variability of each indicator relative to routine soil nutrient tests. Grid soil samples were col-lected three times during a corn (Zea mays L.) growing season at three sites in Ohio. Analytical variability of indicators ranked from highest to lowest mean coefficients of variation (cVs): mineralizable c (13–23%) > OM via loss-on-ignition (OM-LOI; 11–16%) = POXc (9–19%) > protein (2.6–3.2%) = Mehlich-3 P (1.8–2.6%) = Mehlich-3 K (3.0–3.8%) > pH (≤1%). Temporal variability of indicators ranked from highest to lowest mean cVs: miner-alizable c (22–37%) > OM-LOI (16–25%) = POXc (9–21%) = Mehlich-3 P (14–33%) = Mehlich-3 K (6–33%) ≥ protein (7–13%) > pH (1.7–3.6%). Almost all soil properties exhibited moderate to strong spatial autocorrela-tions, occurring at range distances ≤76 m. Our results collectively suggest: (i) mineralizable c and POXc have analytical variability similar to that of OM-LOI, whereas soil protein similar to Mehlich-3 extractable nutrients; (ii) the soil health indicators exhibit temporal variability to a degree similar to routine soil nutrient tests; and (iii) the soil health indicators display spatial variability to a similar extent as the routine soil nutrient tests and therefore do not require greater soil sampling densities within a field.

Abbreviations: AIC, Akaike’s information criterion; BCA, bicinchoninic acid assay; CV, coefficient of variation; LOI, loss-on-ignition; OM, organic matter; POXC, permanganate-oxidizable C.

Soil organic matter (OM) is a key determinant and indicator of both soil fer-tility and soil health (Reeves, 1997; Weil and Magdoff, 2004). Organic mat-ter influences numerous soil functions, including soil structural stability, wa-

ter holding capacity, biological activity, and nutrient retention and release (Tiessen et al., 1994; Weil and Magdoff, 2004). Most standard soil analyses performed by commercial soil testing facilities include the measurement of soil OM (via weight loss on ignition) as part of routine soil nutrient testing program (NCERA-13, 2015). However, soil OM is known to respond very slowly over time to manage-ment changes, and hence may not provide an early indication of potential soil OM accrual or loss (Wander, 2004; Mirsky et al., 2008). Thus, indicators that are more sensitive than total OM are needed to monitor short-term management-induced changes (Doran and Parkin, 1994; Doran and Zeiss, 2000; Doran, 2002; Wienhold et al., 2004; Karlen et al., 1997, 2003, 2008). The soil OM pool includes a continuum of compounds varying in nutrient content and biological availability (Schmidt et al., 2011; Lehmann and Kleber, 2015). The labile OM pool in par-ticular is a good candidate for soil health assessment, as it would be responsive to

Tunsisa T. Hurisso* Steve W. culman Kaiguang Zhao

School of Environ. & Natural Resour. Ohio State Univ. OARDC Wooster, OH 44691

core Ideas

•POXc, mineralizable-c, and protein are emerging soil health indicators.

•POXc and mineralizable-c have analytical variability similar to OM via loss-on-ignition.

•Soil protein has smaller analytical variability similar to Mehlich-3 extractable nutrients.

•Soil health indicators exhibit spatiotemporal variability to a similar extent as routine nutrient tests.

•Soil health indicators do not require greater soil sampling densities within a field.

Published online August 2, 2018

940 Soil Science Society of America Journal

management changes (usually within one to three years of imple-mentation) (Wander, 2004; Weil and Magdoff, 2004; Culman et al., 2013).

Permanganate oxidizable C, mineralizable C, and soil pro-tein are rapid and affordable soil tests that focus on the fast-cy-cling, labile OM pool. A measure of the biologically-available soil C pool, POXC is related to key soil quality indicators including microbial biomass and a processed pool of C (Weil et al., 2003; Mirsky et al., 2008; Culman et al., 2012b). Permanganate oxidiz-able C is also sensitive to management practices (Stine and Weil, 2002; Weil et al., 2003; Culman et al., 2012b; Lucas and Weil, 2012; Culman et al., 2013). Mineralizable C based on short-term aerobic incubation (1 to 3 d) is a general indicator of biological activity (Wang et al., 2003), and is sensitive to management prac-tices (Culman et al., 2013; Ladoni et al., 2015). A comprehensive study showed that POXC and mineralizable C were often cor-related, but differentially influenced by management practices (Hurisso et al., 2016). In that study, POXC was related more to practices that promote soil OM building (e.g., no-till and compost addition) than mineralizable C, which was associated more with practices that lead to soil OM mineralization (e.g., tillage and leguminous cover cropping). Protein-N represents by far the largest fraction of organic-N containing compounds in soil OM, with estimates ranging from 30 to 45% of total soil N (Gillespie et al. (2011) and references therein). Soluble proteins have also been found to be the principal source of mineralizable N (Németh et al., 1988; Matsumoto et al., 2000), indicating that soil proteins can serve as a reservoir of N that is released through mineralization processes ( Jan et al., 2009; Nannipieri and Paul, 2009). Some studies have reported that soil protein is also re-sponsive to management practices, such as tillage and crop ro-tational diversity (Moebius et al., 2007; Moebius-Clune et al., 2008; Roper et al., 2017; Hurisso et al., 2018).

The three measurements described above are also part of indicators included in commercially-available soil health assess-ments, such as the Cornell Assessment of Soil Health Framework (Moebius-Clune et al., 2016). In addition, POXC, mineralizable C and soil protein have been identified as potential soil health in-dicators by recently established initiatives, such as the Soil Health Institute (https://soilhealthinstitute.org/); however, they have yet to be adopted widely by commercial soil testing facilities that han-dle a high volume of samples. These laboratories might be more receptive to incorporating these rapid and inexpensive soil tests into their current portfolios of services if the measurements were easily repeatable (i.e., low analytical variability). Unfortunately, there is no published information on the repeatability of POXC and soil protein, and scant information on mineralizable C repeat-ability, for which the studies by Sherrod et al. (2012) and Wade et al. (2018) are the only reports to our knowledge.

Another key factor to improve the accuracy of soil test results is understanding the spatial and temporal variability associated with soil properties. Total soil C and N are generally stable in time and less variable in space and only affected by long-term changes in soil management (Ruffo et al., 2005). Since POXC, mineralizable

C and soil protein are measures of the labile OM pool, they are more likely to display greater spatiotemporal variability than total soil C and N. Geostatistics are commonly used to assess spatial het-erogeneity of soil properties using model parameters derived from semivariograms, including range distances to describe the extent of spatial autocorrelation (Ettema and Wardle, 2002; Kral et al., 2012). For example, Robertson et al. (1993) found soil organic C, soil pH, and soil test P to be spatially autocorrelated over range distances of 7 to 26 m in an annually-mown and never cropped field. In an adjacent field that was plowed and cropped annually for decades, they determined range distances of 23 to 64 m for the same soil parameters. Their findings demonstrate that the distri-bution of soil properties within a field can be spatially structured over distances of a few meters to hundreds of meters. But pub-lished information on spatial variability of POXC, mineralizable C, and soil protein does not exist. Therefore, a better understand-ing of spatiotemporal variability for these relatively new soil health indicators is needed to help guide farmers and farm advisors on the appropriate sampling densities required to account for inher-ent field spatial variability.

The objectives of this study were: (i) to compare the repeat-ability of POXC, mineralizable C and soil protein measurements with those of routine soil nutrient tests within a single analytical lab, and (ii) to assess in-field spatial and temporal variability of each indicator relative to routine soil nutrient tests. For the pur-pose of this study, routine soil nutrient tests are defined as those recommended by Land Grant Universities (NCERA-13, 2015) and most commonly reported in the U.S. North Central Region (specifically, soil OM via loss of weight on ignition, soil pH, and Mehlich-3 extractable P and K).

MATeRIALS AnD MeTHODSStudy Area, Soil Sampling, and Processing

Soils used in this study were collected from agricultural cropland fields located in Knox, Mahoning and Wood Counties in Ohio (Table 1). The fields were sampled three times over the course of corn growing season in 2016. The first sampling event took place in June and corresponded with soil sampling for pre-sidedress nitrate test in corn, which is typically at V4-V8 growth stage in Ohio (Tremblay et al., 2012), hereafter referred to as V4 sampling for the sake of greater simplicity. The second sampling event occurred at the end of July when corn was at silking or R1 growth stage (hereafter called R1 sampling). The third and fi-nal set of samples were collected in October when corn was at physiological maturity or R6 growth stage (hereafter called R6 sampling). For further details on corn growth and development stages in Ohio, see Thomison et al. (2017). Soils were sampled at different time points during growing season to gain insights into short-term temporal dynamics in measures of the labile OM pool, which is a nutrient pool for both plants and the soil food web, relative to routine soil nutrient tests.

At each field site, soil samples were collected in a grid pattern (46 m by 61 m) to facilitate geostatistical analysis, using semivar-iograms (Kral et al., 2012) to assess spatial autocorrelation in the

www.soils.org/publications/sssaj 941

soil health indicators relative to routine soil nutrient tests. The first sampling area (anchor point) was established 10 m (~12 rows of corn) from the edge of the field and the remaining nineteen sam-pling points were oriented to this anchor point, every 15.2 m apart (Fig. 1). At each center grid point, 12 individual soil cores were randomly collected within a radius of 3.0 to 4.5 m (four to six rows of corn). The soil cores were collected using a probe (2.5 cm i.d.) to a depth of 0- to 20-cm from the middle of the row. All 12 cores from each sampling point in the grid were combined to form a composite soil sample for that grid point. Even though our sam-pling density and number of cores per sample was slightly higher than what is typically practiced in the agriculture industry in Ohio (5 to 8 cores per composited soil sample; ~0.5 to 5 acres per soil sample), we make the assumption in this study that spatial variabil-ity in these fields can be scaled-up to larger areas.

Once in the laboratory, all soil samples were sieved to pass an 8-mm screen, air-dried, and ground to pass through a 2-mm sieve

with a Dynacrush flail grinder, as commonly practiced in most commercial soil testing facilities in the North Central Region (NCERA-13, 2015). All soil samples were analyzed in a single analytical replicate except those from the R6 sampling, which were analyzed in triplicate to evaluate analytical variability. In the case of soil samples analyzed in triplicate, all three replicate subsamples from each composite soil were analyzed at the same time by a single technician within the same analytical laboratory.

Permanganate Oxidizable cAnalysis of POXC was based on the method described

by Weil et al. (2003), with minor modification as described in Culman et al. (2012a). A 20-mL volume of 0.02 mol L–1 KMnO4 solution was added to a 50-mL centrifuge tube contain-ing 2.5 g of soil and shaken for 2 min on a horizontal shaker. The soil was allowed to settle for 10 min, after which 0.5 mL of the supernatant was transferred into a second 50-mL centrifuge tube containing 49.5 mL of deionized water. Sample absorbance values were read at 550 nm with a 96-well spectrophotometric plate reader and calibrated using standard concentration curves of KMnO4 solutions.

Mineralizable cMineralizable C (i.e., C respired upon rewetting of dried soils)

was measured during 1 d of aerobic incubation (Franzluebbers et al., 2000; Franzluebbers, 2016). Permutations of this method exist among researchers, including incubation time (1 to 3 d), sieve size (2–6 mm), and amount of water added. We employed a methodol-ogy amenable to a high-throughput framework. A 10-g soil sample was weighed into a 50-mL polypropylene centrifuge tube. To each tube, deionized water was added to adjust soil water content to 50% water-holding capacity, which was determined based on the differ-ence in weight between a saturated soil that was allowed to drain for an hour and the weight after drying soil overnight in an oven at 105°C. The tubes were capped tightly with lids containing rubber septum and incubated at a room temperature of 23°C for 24 h. The headspace air was mixed by pumping a syringe three times before taking a 1-mL air sample to determine the concentration of CO2

Table 1. Description of Ohio study sites.

county (location) Soil series (taxonomic class) Management practice

Soil texture

Sand Silt clay

---------------------- g kg soil-1 -------------------Knox 40°17¢ N 82°35¢ W

Condit silt loam (Fine, illitic, mesic Typic Epiaqualfs) and Pewamo silty clay loam (Fine, mixed, active, mesic Typic Argiaquolls) with 0 to 2% slopes

Corn (Zea mays L.), soybean [Glycine max (L.) Merr.] and wheat (Triticum aestivum L.) rotation involving tillage before corn planting

260 ± 25† 443 ± 27 297 ± 23

Mahoning40°56¢ N 80°40¢ W

Bogart loam (Fine-loamy, mixed, active, mesic Aquic Hapludalfs) and Jimtown loam (Fine-loamy, mixed, superactive, mesic Aeric Endoaqualfs) with 2 to 6% slopes

Corn-soybean rotation involving chisel plow tillage in the spring before corn planting and broiler chicken (Gallus gallus domesticus) litter application (4.5 Mg ha-1 each spring for over 5 yr) until 2014 when the practice was replaced with mineral P and K fertilizers

423 ± 77 350 ± 74 227 ± 15

Wood41°28¢ N 83°33¢ W

Hoytville clay loam (Fine, illitic, mesic Mollic Epiaqualfs) with 0 to 1% slopes

Corn-soybean-wheat rotation involving tillage in the spring before corn planting and in the fall after corn harvest

471 ± 40 158 ± 52 370 ± 14

† Values are means with one standard deviation.

fig. 1. Soil sampling design for repeatability and spatiotemporal variability assessment of soil properties at three locations in Ohio. The sampling protocol followed a grid pattern which consisted of 20 points spaced at 15.2 m by 15.2 m. A composite soil sample of 12 cores was taken at each sampling point in the grid.

942 Soil Science Society of America Journal

by injecting the air sample into an LI-820 infrared gas analyzer (LI-COR, Biosciences, Lincoln, NE). Mineralizable C was calculated as the difference between a sample and a blank control, using the head-space volume and the ideal gas law (Zibilske, 1994).

Soil ProteinExtractable soil protein was determined using a neutral so-

dium citrate (pH 7) buffer (Hurisso et al., 2018). A 24-mL vol-ume of 0.02 mol L–1 sodium citrate buffer solution was added to 3.0 g of soil contained in a 50-mL glass centrifuge tube. The soil-extractant mixture was shaken for 5 min (180 strokes min-1) and subjected to a high temperature and pressure in an autoclave (121°C, 30 min). The extracts were allowed to cool to room tem-perature, shaken for 3 min (180 strokes min-1), and then clarified by centrifugation (10,000 g, 3 min). The quantity of extracted protein was measured using the colorimetric bicinchoninic-acid (BCA) assay (Thermo Scientific, Pierce, Rockford, IL) with a 96-well spectrophotometric plate reader at 562 nm. Sample ab-sorbance readings were calibrated using a standard curve of 0 to 2000 mg mL-1 bovine serum albumin. Autoclaved-citrate extract-able soil protein content of each sample was calculated by mul-tiplying protein concentration in soil extracts by the volume of exractant used, and dividing the results by the weight of soil used.

Routine Soil nutrient TestsAll routine soil analyses followed the procedures outlined

and recommended by Land Grant Universities in the North Central Region (NCERA-13, 2015). Soil OM was determined by loss of weight on ignition (OM-LOI) in a high temperature oven at 360°C for 2 h (Combs and Nathan, 1998). Soil water pH was measured with a glass electrode in a 1:1 soil/water (w/v) mixture (Peters et al., 2012). Extractable soil P and K were de-termined using the Mehlich-3 extractant (Mehlich, 1984) and analyzed with an inductively coupled plasma spectrometer as de-scribed by Frank et al. (1998) and Warncke and Brown (1998). In addition, particle size was determined by the standard hy-drometer method according to Gee and Bauder (1986).

Statistical AnalysesTo estimate the relative magnitude of analytical and tempo-

ral variability associated with each soil test, coefficient of varia-tion (CV) was calculated as standard deviation normalized by the mean, separately for each grid point (n = 20). For analytical variability, the CV calculations were based on measurements per-formed on three replicate subsamples of the same composite soil at each grid point from R6 sampling. The CV calculations for temporal variability were based on measurements performed on a single replicate sample at each grid point from V4, R1, and R6 sampling times. The CV value was then used as a response vari-able in analysis of variance model where soil tests and grid points were treated as fixed and random effects, respectively. Data was checked for parametric statistical assumptions and analyzed by using the PROC MIXED procedure in SAS (SAS ver. 9.4, SAS Institute Inc., Cary, NC). Differences between means were sepa-

rated using the PDIFF option of the LSMEANS statement with Tukey’s HSD adjustment at p < 0.05 unless reported otherwise. All graphs were made with the function ggplot() from the gll-plot2 package (RStudio Team, 2016).

The presence of spatial autocorrelations was first assessed by calculating and plotting sample semivariograms of the dataset by using custom R codes (RStudio Team, 2016). The semivariance calculation was according to the formula:

( ) [ ]1

1 ( ) ( )

2

hN

i iih

h Z s Z s hN

g=

= - +∑where Nh is the number of observation pairs separated by a dis-tance of h, Z(si) is the value of the variable of interest at loca-tion si, and Z(si + h) is the value of that same variable of interest at a location at distance h from si. Then theoretical semivario-gram models that account for spatial correlations, including exponential, Gaussian, power and spherical were fitted to the data, separately for each site and sampling time. The Akaike Information Criterion (AIC) was used to compare model per-formances. Spatial autocorrelation was considered to be present when AIC of at least one of the semivariogram models that ac-counted for spatial correlations was lower than AIC from the null model that assumed no spatial autocorrelation (Burnham et al., 2011; Symonds and Moussalli, 2011). Range distances de-rived from the best-fit model were used to describe the extent of spatial dependence. The proportion of model sample variance explained by structural variance [C/(C0 + C)] was calculated as a normalized measure of spatial dependence, using nugget (C0) and the sill variance (C0 + C) from best-fit models. The values of [C/(C0 + C)] range from 0 to 1, where [C/(C0 + C)] < 0.25, 0.25 < [C/(C0 + C)] < 0.75, and [C/(C0 + C)] > 0.75 respec-tively indicate weak, moderate, and strong spatial dependence (Robertson et al., 1993; Cambardella et al., 1994). For the pur-pose of spatial and temporal variability assessment, we used only one analytical replicate from the R6 soils that were analyzed in triplicate by randomly selecting one replicate using the function sample_n() from the dplyr package.

Pearson’s correlation coefficients were also calculated using the function rcorr() from the Hmisc package to assess the relation-ships among the different soil tests across sites and sampling times. Significant correlations were identified at p < 0.05 or p < 0.10.

ReSULTS AnD DIScUSSIOnDescriptive Statistics

Across sites, measured soil POXC ranged from 93 to 991 mg kg-1, mineralizable C ranged from 7 to 95 mg kg-1 and soil protein ranged from 3.6 to 9.6 mg g-1 (Table 2). These values are within the ranges of POXC, mineralizable C and soil protein values found in the literature (Culman et al., 2012b; Hurisso et al., 2016; Fine et al., 2017). Wood County generally had larger values of organic matter (POXC, mineralizable C, soil protein, OM-LOI), likely reflecting the higher percentage of clay relative to the other sites (Tables 1 and 2).

www.soils.org/publications/sssaj 943

Soil health indicators were generally positively related to routine soil measurements, but the strength of the relationship varied considerably (Table 3). In particular, OM-LOI was cor-related with POXC, mineralizable C and soil protein. The range of r2 values of the correlations between OM-LOI and POXC, mineralizable C, and soil protein were 0.20 to 0.54, 0.19 to 0.36, and 0.24 to 0.61 respectively, depending on site and sampling time (data not shown). Consistent with results presented here, Fine et al. (2017) also reported significant relationships of OM-LOI with POXC (r2 = 0.52), mineralizable C (r2 = 0.61), and soil protein (r2 = 0.45) for samples collected from a wide range of soils and cropping systems in the Mid-Atlantic, Midwest, and Northeast regions.

Analytical Variability

Soil testing laboratories require soil tests with relatively small analytical variability so that meaningful data can be generated consistently from samples submitted for analysis. Thus, a primary question of interest when adopting new methodology is to what extent are emerging soil health indicators repeatable? This ques-tion focuses on the precision of these relatively new indicators that are ideal for routine soil health evaluation, using CV values from a group of laboratory replicates of the same composited soil sample to assess analytical variability. Across all sites, POXC, mineraliz-able C and soil protein mean CV values ranged from 9 to 19, 13 to 23 and 2.6 to 3.2%, respectively. Trends in the analytical variability of each indicator were largely consistent across sites, with organic matter measurements CV values following the pattern: mineraliz-

Table 2. Descriptive statistics for the various soil tests by site and time of sampling (n = 20 for each mean value).

Knox Mahoning WoodStatistic V4 R1 R6 V4 R1 R6 V4 R1 R6

POXc†-------------------------------------------------------------------------------------------- mg kg soil-1 ---------------------------------------------------------------------------------------------

Mean 456 491 422 493 459 395 638 668 724Std. Dev. 86 88 91 68 127 167 72 61 78Minimum 319 319 290 406 318 94 437 511 589Maximum 657 656 612 662 887 991 736 769 908

Mineralizable c-------------------------------------------------------------------------------------------- mg kg soil-1 --------------------------------------------------------------------------------------------

Mean 50 52 39 41 33 22 44 46 40Std. Dev. 15 16 14 18 19 7 11 8 7Minimum 21 33 11 19 12 7 22 24 28Maximum 76 91 61 95 90 41 60 58 55

Protein-------------------------------------------------------------------------------------------- mg kg soil-1 --------------------------------------------------------------------------------------------

Mean 5.6 5.1 5.0 6.7 6.2 5.3 6.5 6.5 7.0Std. Dev. 0.6 0.5 0.7 0.9 1.0 0.7 0.7 0.5 0.4Minimum 4.5 4.4 3.6 5.2 5.1 4.0 5.7 5.9 6.2Maximum 6.7 6.0 6.0 9.0 9.6 7.0 7.6 7.6 7.7

OM-LOI‡---------------------------------------------------------------------------------------------- g kg soil-1 ----------------------------------------------------------------------------------------------

Mean 25 25 22 23 23 15 29 29 33Std. Dev. 6 5 7 3 4 3 3 4 9Minimum 16 18 10 18 18 9 25 21 3Maximum 40 35 33 29 37 22 34 37 54

pHMean 5.6 5.6 5.9 5.7 5.6 5.7 6.3 6.2 6.3Std. Dev. 0.4 0.3 0.4 0.3 0.5 0.4 0.1 0.2 0.3Minimum 5.0 5.1 5.3 5.2 5.1 5.1 6.0 5.9 5.8Maximum 6.5 6.1 6.4 6.2 6.8 6.5 6.5 6.7 7.1

Mehlich-3 P-------------------------------------------------------------------------------------------- mg kg soil-1 --------------------------------------------------------------------------------------------

Mean 112 125 105 79 83 44 61 66 78Std. Dev. 36 44 31 43 58 28 8 13 15Minimum 63 64 49 44 40 17 49 44 52Maximum 203 231 184 200 275 150 78 89 105

Mehlich-3 K-------------------------------------------------------------------------------------------- mg kg soil-1 --------------------------------------------------------------------------------------------

Mean 187 170 157 351 339 189 208 204 224Std. Dev. 34 35 36 69 123 43 9 13 11Minimum 124 96 85 214 199 123 189 183 202Maximum 259 237 218 454 727 280 225 230 243† POXC, permanganate-oxidizable carbon.‡ OM-LOI, soil organic matter determined via loss-on-ignition.

944 Soil Science Society of America Journal

able C > OM-LOI = POXC > soil protein (Fig. 2). Soil pH had the lowest analytical variability, likely given its logarithmic scale, and Mehlich-3 P and K were intermediate in variability.

These results demonstrate that analytical variability associ-ated with POXC and mineralizable C are on a similar scale to OM-LOI, but soil protein has at least fourfold smaller variability compared with the other organic matter measurements (POXC,

mineralizable C, and OM-LOI). The relatively large analytical variability associated with POXC and mineralizable C will need to be considered in routine soil testing. Since high analytical vari-ability associated with a soil test can obscure differences among imposed management practices, measurements of OM-LOI, POXC and mineralizable C may need to be run with analytical replication. In particular, mineralizable C is known to be inher-

ently a variable measurement (Ahn et al., 2009; Zagal et al., 2009; Morrow et al., 2016; Wade et al., 2018). Consistent with our find-ings, Wade et al. (2018) also found a considerable amount of analytical variability in mineralizable C with-in a laboratory, which was highly dependent on the type of soils they studied. Although our results in-dicated similarity between POXC and OM-LOI in terms of analyti-cal variability, it is worthwhile to mention that the former has an advantage of being more respon-sive to changes in soil management than OM evaluated on a total soil C basis (Weil et al., 2003; Culman et al., 2012b, 2013; Hurisso et al., 2016; Morrow et al., 2016).

Temporal VariabilityAcross all three sites, tempo-

ral trends were also evident, but were different depending on the site and indicator. At the Knox and Mahoning sites, soil health and soil nutrient measures typically were highest at the V4 or R1 sampling and declined at the R6 sampling (Table 2). Whereas at the Wood site, most soil health and soil nu-trient indicators were greatest at the R6 sampling. The soil at the

Table 3. Pearson correlation coefficients (r) and p-values for soil health and soil nutrient tests calculated across all sites and sam-pling times (n = 180).

POXc† Mineralizable c Protein OM-LOI‡ pH Mehlich-3 P Mehlich-3 KPOXC 1.000 *** *** *** *** NS§ *Mineralizable C 0.464 1.000 *** *** *** *** NSProtein 0.703 0.365 1.000 *** *** NS ***OM-LOI 0.627 0.451 0.511 1.000 *** ¶ ¶pH 0.695 0.455 0.460 0.491 1.000 ¶ NSMehlich-3 P 0.005 0.407 0.093 0.132 -0.135 1.000 *Mehlich-3 K 0.167 0.123 0.573 0.139 0.054 0.155 1.000* Significant at p < 0.05; *** Significant at p < 0.001. † POXC, permanganate-oxidizable carbon. ‡ OM-LOI, soil organic matter determined via loss-on-ignition. § NS, not statistically significant.¶ Significant at p < 0.10.

fig. 2. Analytical variability, expressed as coefficients of variation (cV), associated with emerging soil health indicators and routine soil nutrient tests. each bar represents the mean (n = 20) of cV values calculated based on measurements performed on a group of three replicate subsamples from the same composite soil from the R6 sampling. Within a site, mean cV values followed by different lowercase letters were significantly different at p < 0.05. error bars represent the standard error of the mean. POXc, permanganate-oxidizable carbon; LOI, loss-on-ignition.

www.soils.org/publications/sssaj 945

Wood site has more clay than soils at Knox and Mahoning sites (Table 1), which may have contributed to these differential effects. The temporal patterns observed in POXC and mineralizable C at the Knox site were similar to those reported by Culman et al. (2013), peaking at R1 sampling; however, the values peaked at V4 sampling at the Mahoning site and R6 sampling at the Wood site. Such inconsistent patterns in the mean values measured for each soil health indicator and soil nutrient test (Table 2) could be due to variations in uptake and losses of nutrients and addition of labile OM via rhizodeposition and root turnover during the grow-ing season.

To further determine the extent to which the soil health indicators vary temporally relative to routine soil nutrient tests, we computed CV values using measurements performed on soils from each sampling time points. Patterns in the magnitude of temporal variability of each indicator were less consistent across sites, but mean CV values for organic matter measurements fol-lowed the order: mineralizable C (22 to 37%) ≥ OM-LOI (16 to 25%) ≥ POXC (9 to 21%) = soil protein (7 to 13%; Fig. 3). Soil pH had consistently lower tempo-ral variability than the soil health indicators considered in this study except soil protein. Whereas Mehlich-3 P and K were either statistically similar or higher in temporal variability than POXC and soil protein, with their respec-tive mean CV values ranging from 14 to 33% and 6 to 33% (Fig. 3). Overall, data presented here dem-onstrate the tendency for both the soil health indicators and rou-tine soil nutrient tests to change considerably over the course of a growing season. Thus, these results underscore the need to collect soil samples at the same time of the year, as is usually recommended due to known variability in routine soil test measurements.

Spatial VariabilityAnother key question we

addressed was to what extent do these emerging soil health indica-tors vary spatially relative to rou-tine soil nutrient tests. Exploring this question could help provide very useful information for design-ing appropriate soil sampling strat-egies needed to address in-field spatial variability (e.g., whether soil cores should be taken every 10 m or every 1000 m across a field).

Almost all soil properties examined here exhibited moderate to strong spatial dependence, with the proportion of model sample variance explained by structural variance, that is, C/(C0 + C), ranging from 0.26 to 0.98 depending on site and sampling time (Table 4). At the Wood site, most soil tests (in particular soil nutrient tests) did not exhibit spatial autocorrelation. The dis-cussion regarding spatial autocorrelations and range distances is therefore focused on the Knox and Mahoning sites unless speci-fied otherwise. Spatial range distances differed between sites and among sampling times for most indicators, but not consistently. At the Knox and Mahoning sites, POXC was spatially autocor-related over a distance ranging from 43 to 76 m (Table 4). For mineralizable C, the range distances were between 23 and 76 m, whereas range distances of 31 to 76 m were determined for soil protein across sites and sampling times (Table 4).

Similar inconsistent spatial autocorrelation trends were found for most routine soil nutrient tests as for the soil health in-dicators. The sole exception was soil test P, which was consistent-ly autocorrelated to a spatial distance of 76 m independent of site

fig. 3. Temporal variability, expressed as coefficients of variations (cV), associated with emerging soil health indicators and routine soil nutrient tests. each bar represents the mean (n = 20) of cV values calculated based on measurements performed on a single replicate sample at each grid point from V4, R1 and R6 samplings. Within a site, mean cV values followed by different lowercase letters were significantly different at p < 0.05. error bars represent the standard error of the mean. POXc, permanganate oxidizable carbon; LOI, loss-on-ignition.

946 Soil Science Society of America Journal

and sampling time (Table 4). Other researchers have also report-ed spatial autocorrelations for soil P over a distance ranging from 64 to 78 m in corn-based cropping systems (Robertson et al., 1993; Cambardella et al., 1994; Cambardella and Karlen, 1999). The OM-LOI range distance was between 10 and 76 m across sites and sampling times (Table 4), which is also consistent with values found in the literature. Maresma and Ketterings (2017) determined spatial range distance of 38 to 79 m for OM-LOI in corn fields, depending on site and sampling season, and Baxter et al. (2003) reported OM-LOI range distance of 49 m for a field under winter barley (Hordeum vulgare L.) production for 10 yr. For total soil C (determined via dry combustion), Cambardella et al. (1994) reported spatial autocorrelation over a distance of 104 m, which is bigger than the spatial ranges for OM-LOI in our study and those of others (Baxter et al., 2003; Maresma and Ketterings, 2017). Such differences in spatial ranges could be

due to differences in field management history (Robertson et al., 1993; Cambardella and Karlen, 1999).

These results collectively demonstrate that the distribution of the soil health indicators can be spatially autocorrelated over distances of few meters to 76 m depending on site and sampling time. The fact that spatial autocorrelations occurred at range distances ≤76 m for both the soil health indicators and routine soil nutrient tests (Table 3) suggests that the three soil health indicators investigated here exhibit in-field spatial variability to a similar extent as the routine soil nutrient tests. Therefore, the sampling densities typically used for routine soil nutrient analysis (a composite soil sample of 5 to 8 cores with each composite soil sample representing ~0.5 to 5 acres), with cores taken approxi-mately 76 m apart, should also be adequate for the measurement of POXC, mineralizable C and soil protein.

Table 4. Spatial dependence and spatial range from best-fit semivariograms for emerging soil health indicators and routine soil nutrient tests by site and time of sampling.†

Site

V4 R1 R6

Model‡Spatial

dependence§ Range ModelSpatial

dependence Range ModelSpatial

dependence Range

m m mPOXc

Knox Sph 0.81 76 Sph 0.87 76 Exp 0.68 47

Mahoning Sph 0.87 43 Gau 0.91 76 Sph 0.98 76

Wood Exp 0.80 42 –¶ – – Sph 0.90 76

Mineralizable c

Knox Exp 0.78 76 Exp 0.68 76 Sph 1.00 23

Mahoning Gau 0.89 76 Gau 0.89 76 – – –

Wood Exp 0.95 76 – – – Exp 0.26 17

Protein

Knox Sph 0.90 31 Exp 0.59 76 Sph 1.02 32

Mahoning Sph 0.44 44 Exp 0.98 76 Exp 0.46 76

Wood – – – – – – Exp 0.67 76

OM-LOI

Knox Sph 1.32 22 Exp 0.69 10 Sph 1.13 76

Mahoning Gau 0.52 10 Sph 0.68 76 Sph 0.76 26

Wood – – – – – – – – –

pH

Knox Gau 0.98 76 Sph 1.09 76 Sph 1.72 24

Mahoning Gau 0.94 36 Gau 0.92 76 Gau 0.94 28

Wood – – – – – – – – –

Mehlich-3 P

Knox Gau 0.94 76 Gau 0.92 43 Gau 0.84 76

Mahoning Exp 0.87 76 Exp 0.97 76 Gau 0.92 76

Wood – – – – – – Gau 0.91 32

Mehlich-3 K

Knox Sph 1.00 36 Sph 1.00 28 Sph 1.00 54

Mahoning Sph 1.00 23 Sph 1.00 76 Sph 0.67 76Wood – – – Exp 1.00 59 Sph 0.87 31† POXC, permanganate-oxidizable carbon; OM-LOI, soil organic matter determined via loss-on-ignition.‡ Model semivariograms: Exp = Exponential; Gau = Gaussian; Sph = Spherical.§ Spatial dependence = [C/(C0+C)], where (C0) and (C0 + C) respectively represent the nugget and the sill variance. The higher the value the

greater the strength of spatial correlation over the range of the separation distance.¶ Negative sign (–) indicates lack of spatial dependence (see the statistical analysis section for more details as to how the presence vs. absence of

spatial dependence was determined).

www.soils.org/publications/sssaj 947

cOncLUSIOnThis study examined repeatability and spatiotemporal vari-

ability of emerging soil health indicators relative to routine soil nutrient tests that are commonly used in the US North Central Region. Among the soil health indicators, the analytical vari-ability of POXC and mineralizable C was relatively large and on a similar scale to that of OM-LOI. In contrast, soil protein was found to be analytically the least variable and similar to Mehlich-3 extractable nutrients, suggesting that measurements of soil protein should be easily repeatable in a soil testing labo-ratory. Our results also suggested that the soil health indicators considered in this study can vary temporally during a growing season to a similar degree as routine soil nutrient tests; as a conse-quence, soil samples for measuring POXC, mineralizable C and soil protein should likely be taken at the same time of the year, as is typically recommended in routine soil nutrient testing. Data presented here also showed that both the soil health indicators and routine soil nutrient tests exhibited spatial autocorrelations over similar range distances of ≤76 m, suggesting that measure-ments of POXC, mineralizable C and soil protein should not require greater soil sampling densities within a field than what is typically used for routine soil nutrient testing.

AcKnOWLeDGMenTSThe authors are grateful for Alan Sundemier, John Barker, Lee Beers, Skyler Foos, Josh Isaacson and Evan Schaefer for their assistance with study site identification and initial soil sampling. We are also grateful for the technical support from Phoo Zone, Stuti Sharma, and Bethany Herman during soil sampling and laboratory analyses. This work was supported by the Organic Agriculture Research & Extension Initiative (2014-51300-22331) from the USDA National Institute of Food and Agriculture. We would also like to thank three anonymous reviewers for providing thoughtful comments on the paper.

RefeRenceSAhn, M.-Y., A.R. Zimmerman, N.B. Comerford, J.O. Sickman, and S. Grunwald.

2009. Carbon mineralization and labile organic carbon pools in the sandy soils of a North Florida watershed. Ecosystems 12:672–685. doi:10.1007/s10021-009-9250-8

Baxter, S.J., M.A. Oliver, and J. Gaunt. 2003. A geostatistical analysis of the spatial variation of soil mineral nitrogen and potentially available nitrogen within an arable field. Precis. Agric. 4:213–226. doi:10.1023/A:1024565507688

Burnham, K.P., D.R. Anderson, and K.P. Huyvaert. 2011. AIC model selection and multimodel inference in behavioral ecology: Some background, observations, and comparisons. Behav. Ecol. Sociobiol. 65:23–35. doi:10.1007/s00265-010-1029-6

Cambardella, C.A., and D.L. Karlen. 1999. Spatial analysis of soil fertility parameters. Precis. Agric. 1:5–14. doi:10.1023/A:1009925919134

Cambardella, C.A., T.B. Moorman, J.M. Novak, T.B. Parkin, D.L. Karlen, R.F. Turco, and A.E. Konopka. 1994. Field-scale variability of soil properties in central Iowa soils. Soil Sci. Soc. Am. J. 58:1501–1511. doi:10.2136/sssaj1994.03615995005800050033x

Combs, M., and M.V. Nathan. 1998. Soil organic matter. In: M. Nathan, and R. Gelderman, editors, Recommended chemical soil test procedures for the North Central Region. North Central Regional Research Publication No. 221 (Revised 2015). Missouri Agricultural Experiment Station SB 1001. Chap. 12. Univ. of Missouri, Columbia, MO.

Culman, S.W., M. Freeman, and S.S. Snapp. 2012a. Procedure for the determination of permanganate oxidizable carbon. Kellogg Biological Station-Long Term Ecological Research Protocols, Hickory Corners, MI. http://lter.kbs.msu.edu/protocols/133.

Culman, S.W., S.S. Snapp, M.A. Freeman, M.E. Schipanski, J. Beniston, R.

Lal, L.E. Drinkwater, A.J. Franzluebbers, J.D. Glover, A.S. Grandy, J. Lee, J. Six, J.E. Maul, S.B. Mirksy, J.T. Spargo, and M.M. Wander. 2012b. Permanganate oxidizable carbon reflects a processed soil fraction that is sensitive to management. Soil Sci. Soc. Am. J. 76:494–504. doi:10.2136/sssaj2011.0286

Culman, S.W., S.S. Snapp, J.M. Green, and L.E. Gentry. 2013. Short- and long-term labile soil carbon and nitrogen dynamics reflect management and predict corn agronomic performance. Agron. J. 105:493–502. doi:10.2134/agronj2012.0382

Doran, J.W. 2002. Soil health and global sustainability: Translating science into practice. Agric. Ecosyst. Environ. 88:119–127. doi:10.1016/S0167-8809(01)00246-8

Doran, J.W., and T.B. Parkin. 1994. Defining and assessing soil quality. In: J.W. Doran, J.A.E. Molina, and R.F. Harris, editors, Defining soil quality for a sustainable environment. SSSA Publ. 35. SSSA, Madison, WI. p. 3–21. doi:10.2136/sssaspecpub35.c1

Doran, J.W., and M.R. Zeiss. 2000. Soil health and sustainability: Managing the biotic component of soil quality. Appl. Soil Ecol. 15:3–11. doi:10.1016/S0929-1393(00)00067-6

Ettema, C.H., and D.A. Wardle. 2002. Spatial soil ecology. Trends Ecol. Evol. 17:177–183. doi:10.1016/S0169-5347(02)02496-5

Fine, A., H. Van Es, and R. Schindelbeck. 2017. Statistics, scoring functions, and regional analysis of a comprehensive soil health database. Soil Sci. Soc. Am. J. 81:589–601. doi:10.2136/sssaj2016.09.0286

Frank, K., D. Beegle, and J. Denning. 1998. Phosphorus. In: M. Nathan and R. Gelderman, editors, Recommended chemical soil test procedures for the North Central Region. North Central Regional Research Publication No. 221 (Revised 2015). Missouri Agricultural Experiment Station SB 1001. Chap. 6. Univ. of Missouri, Columbia, MO.

Franzluebbers, A.J. 2016. Should soil testing services measure soil biological activity? Agric. Environ. Lett. 1:150009. doi:10.2134/ael2015.11.0009

Franzluebbers, A.J., R.L. Haney, C.W. Honeycutt, H.H. Schomberg, and F.M. Hons. 2000. Flush of carbon dioxide following rewetting of dried soil relates to active organic pools. Soil Sci. Soc. Am. J. 64:613–623. doi:10.2136/sssaj2000.642613x

Gee, G.W., and J.W. Bauder. 1986. Particle-size analysis. In: A. Klute, editor, Methods of soil analysis. Part 1. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI. p. 383–411.

Gillespie, A.W., R.E. Farrell, F.L. Walley, A.R.S. Ross, P. Leinweber, K.U. Eckhardt, T.Z. Regier, and R.I.R. Blyth. 2011. Glomalin-related soil protein contains non-mycorrhizal-related heat-stable proteins, lipids and humic materials. Soil Biol. Biochem. 43:766–777. doi:10.1016/j.soilbio.2010.12.010

Hurisso, T.T., S.W. Culman, W.R. Horwath, J. Wade, D. Cass, J.W. Beniston, T.M. Bowles, A.S. Grandy, A.J. Franzluebbers, M.E. Schipanski, S.T. Lucas, and C.M. Ugarte. 2016. Comparison of Permanganate-Oxidizable Carbon and Mineralizable Carbon for Assessment of Organic Matter Stabilization and Mineralization. Soil Sci. Soc. Am. J. 80:1352–1364. doi:10.2136/sssaj2016.04.0106

Hurisso, T.T., D.J. Moebius-Clune, S.W. Culman, B.N. Moebius-Clune, J. Thies, and H.M. van Es. 2018. Soil protein as a rapid soil health indicator of potentially available organic nitrogen. Agric. Environ. Lett. 3:180006. doi:10.2134/ael2018.02.0006

Jan, M.T., P. Roberts, S.K. Tonheim, and D.L. Jones. 2009. Protein breakdown represents a major bottleneck in nitrogen cycling in grassland soils. Soil Biol. Biochem. 41:2272–2282. doi:10.1016/j.soilbio.2009.08.013

Karlen, D., M. Mausbach, J. Doran, R. Cline, R. Harris, and G. Schuman. 1997. Soil quality: A concept, definition, and framework for evaluation (a guest editorial). Soil Sci. Soc. Am. J. 61:4–10. doi:10.2136/sssaj1997.03615995006100010001x

Karlen, D.L., S.S. Andrews, B.J. Weinhold, and T.M. Zobeck. 2008. Soil quality assessment: Past, present and future. J. Integr. Biosci. 6:3–14.

Karlen, D.L., C.A. Ditzler, and S.S. Andrews. 2003. Soil quality: Why and how? Geoderma 114:145–156. doi:10.1016/S0016-7061(03)00039-9

Kral, F., R. Corstanje, J.R. White, and F. Veronesi. 2012. A geostatistical analysis of soil properties in the Davis Pond Mississippi freshwater diversion. Soil Sci. Soc. Am. J. 76:1107–1118. doi:10.2136/sssaj2011.0206

Ladoni, M., A. Basir, and A. Kravchenko. 2015. Which soil carbon fraction is the best for assessing management differences? A statistical power perspective. Soil Sci. Soc. Am. J. 79:848–857. doi:10.2136/sssaj2014.10.0426

948 Soil Science Society of America Journal

Lehmann, J., and M. Kleber. 2015. The contentious nature of soil organic matter. Nature 528:60–68.

Lucas, S.T., and R.R. Weil. 2012. Can a labile carbon test be used to predict crop responses to improved soil organic matter management? Agron. J. 104:1160–1170. doi:10.2134/agronj2011.0415

Maresma, A., and Q.M. Ketterings. 2017. In-field variability of Illinois soil nitrogen test and loss-on-ignition for nitrogen management. Soil Sci. Soc. Am. J. 81:1211–1221. doi:10.2136/sssaj2016.10.0334

Matsumoto, S., N. Ae, and M. Yamagata. 2000. Extraction of mineralizable organic nitrogen from soils by a neutral phosphate buffer solution. Soil Biol. Biochem. 32:1293–1299. doi:10.1016/S0038-0717(00)00049-3

Mehlich, A. 1984. Mehlich 3 soil test extractant: A modification of the Mehlich 2 extractant. Commun. Soil Sci. Plant Anal. 15:1409–1416. doi:10.1080/00103628409367568

Mirsky, S.B., L.E. Lanyon, and B.A. Needelman. 2008. Evaluating soil management using particulate and chemically labile soil organic matter fractions. Soil Sci. Soc. Am. J. 72:180–185. doi:10.2136/sssaj2005.0279

Moebius, B.N., H.M. van Es, R.R. Schindelbeck, O.J. Idowu, J.E. Thies, and D.J. Clune. 2007. Evaluation of laboratory-measured soil properties as indicators of soil physical quality. Soil Sci. 172:895–912. doi:10.1097/ss.0b013e318154b520

Moebius-Clune, B.N., D.J. Moebius-Clune, B.K. Gugino, O.J. Idowu, R.R. Schindelbeck, A.J. Ristow, H.M. van Es, J.E. Thies, H.A. Shayler, M.B. McBride, D.W. Wolfe, and G.S. Abawi. 2016. Comprehensive assessment of soil health–Cornell Framework Manual, ed. 3.1. Cornell Univ., Geneva, NY.

Moebius-Clune, B.N., H.M. van Es, O.J. Idowu, R.R. Schindelbeck, D.J. Moebius-Clune, D.W. Wolfe, G.S. Abawi, J.E. Thies, B.K. Gugino, and R. Lucey. 2008. Long-term effects of harvesting maize stover and tillage on soil quality. Soil Sci. Soc. Am. J. 72:960–969. doi:10.2136/sssaj2007.0248

Morrow, J.G., D.R. Huggins, L.A. Carpenter-Boggs, and J.P. Reganold. 2016. Evaluating measures to assess soil health in long-term agroecosystem trials. Soil Sci. Soc. Am. J. 80:450–462. doi:10.2136/sssaj2015.08.0308

Nannipieri, P., and E. Paul. 2009. The chemical and functional characterization of soil N and its biotic components. Soil Biol. Biochem. 41:2357–2369. doi:10.1016/j.soilbio.2009.07.013

NCERA-13. 2015. Recommended chemical soil test procedures for the North Central Region. North Central Regional Research Publication No. 221. Missouri Agricultural Experiment Station SB 1001. Univ. of Missouri, Columbia, MO. http://extension.missouri.edu/explorepdf/specialb/sb1001.pdf.

Németh, K., H. Bartels, M. Vogel, and K. Mengel. 1988. Organic nitrogen compounds extracted from arable and forest soils by electro-ultrafiltration and recovery rates of amino-acids. Biol. Fertil. Soils 5:271–275. doi:10.1007/BF00262130

Peters, J.B., M.V. Nathan, and C.A.M. Laboski. 2012. pH and lime requirement. Ch. 4. In: M. Nathan, and R. Gelderman, editors, Recommended chemical soil test procedures for the North Central Region. North Central Regional Research Publication No. 221 (Revised 2015). Missouri Agricultural Experiment Station SB 1001. Univ. of Missouri, Columbia, MO.

RStudio Team. 2016. RStudio: Integrated development for R. RStudio, Inc., Boston, MA.

Reeves, D.W. 1997. The role of soil organic matter in maintaining soil quality in continuous cropping systems. Soil Tillage Res. 43:131–167. doi:10.1016/S0167-1987(97)00038-X

Robertson, G.P., J.R. Crum, and B.G. Ellis. 1993. The spatial variability of soil resources following long-term disturbance. Oecologia 96:451–456. doi:10.1007/BF00320501

Roper, W.R., D.L. Osmond, J.L. Heitman, M.G. Wagger, and S.C. Reberg-Horton. 2017. Soil health indicators do not differentiate among agronomic management systems in North Carolina soils. Soil Sci. Soc. Am. J. 81:828–

843. doi:10.2136/sssaj2016.12.0400Ruffo, M.L., G.A. Bollero, R.G. Hoeft, and D.G. Bullock. 2005. Spatial

variability of the Illinois Soil Nitrogen Test: Implications for soil sampling. Agron. J. 97:1485–1492. doi:10.2134/agronj2004.0323

Schmidt, M.W.I., M.S. Torn, S. Abiven, T. Dittmar, G. Guggenberger, I.A. Janssens, M. Kleber, et al. 2011. Persistence of soil organic matter as an ecosystem property. Nature 478:49–56.

Sherrod, L.A., J.D. Reeder, W. Hunter, and L.R. Ahuja. 2012. Rapid and cost-effective method for soil carbon mineralization in static laboratory incubations. Commun. Soil Sci. Plant Anal. 43:958–972. doi:10.1080/00103624.2012.653031

Stine, M.A., and R.R. Weil. 2002. The relationship between soil quality and crop productivity across three tillage systems in south central Honduras. Am. J. Altern. Agric. 17:2–8.

Symonds, M.R.E., and A. Moussalli. 2011. A brief guide to model selection, multimodel inference and model averaging in behavioural ecology using Akaike’s information criterion. Behav. Ecol. Sociobiol. 65:13–21. doi:10.1007/s00265-010-1037-6

Thomison, P., A. Michel, K. Tilmon, S. Culman, and P. Paul. 2017. Ch. 4. Corn production. Ohio agronomy guide. 15th ed. Ext. Bull. 472. Ohio State Univ., Columbus, OH. p. 32–54.

Tiessen, H., E. Cuevas, and P. Chacon. 1994. The role of soil organic matter in sustaining soil fertility. Nature 371:783–785. doi:10.1038/371783a0

Tremblay, N., Y.M. Bouroubi, C. Bélec, R.W. Mullen, N.R. Kitchen, W.E. Thomason, S. Ebelhar, D.B. Mengel, W.R. Raun, D.D. Francis, E.D. Vories, and I. Ortiz-Monasterio. 2012. Corn response to nitrogen is influenced by soil texture and weather. Agron. J. 104:1658–1671. doi:10.2134/agronj2012.0184

Wade, J., S.W. Culman, T.T. Hurisso, R.O. Miller, L. Baker, and W.R. Horwath. 2018. Sources of variability that compromise mineralizable carbon as a soil health indicator. Soil Sci. Soc. Am. J. 82:243–252. doi:10.2136/sssaj2017.03.0105

Wander, M. 2004. Soil organic matter fractions and their relevance to soil function. In: F. Magdoff, and R. Weil, editors, Soil organic matter in sustainable agriculture. CRC Press, Boca Raton, FL. doi:10.1201/9780203496374.ch3

Wang, W.G., R.C. Dalal, P.W. Moody, and C.J. Smith. 2003. Relationships of soil respiration to microbial biomass, substrate availability and clay content. Soil Biol. Biochem. 35:273–284. doi:10.1016/S0038-0717(02)00274-2

Warncke, D., and J.R. Brown. 1998. Potassium and other basic cations. In: M. Nathan, and R. Gelderman, editors, Recommended chemical soil test procedures for the North Central Region. North Central Regional Research Publication No. 221 (Revised 2015). Missouri Agricultural Experiment Station SB 1001. Chap. 7. Univ. of Missouri, Columbia, MO.

Weil, R., and F. Magdoff. 2004. Significance of soil organic matter to soil quality and health. In: F. Magdoff, and R. Weil, editors, Soil organic matter in sustainable agriculture. CRC Press, Boca Raton, FL. doi:10.1201/9780203496374.ch1

Weil, R.R., K.R. Islam, M.A. Stine, J.B. Gruver, and S.E. Samson-Liebig. 2003. Estimating active carbon for soil quality assessment: A simplified method for laboratory and field use. Am. J. Alternative Agric. 18:3–17. doi:10.1079/AJAA2003003

Wienhold, B.J., S.S. Andrews, and D.L. Karlen. 2004. Soil quality: a review of the science and experiences in the USA. Environmental Geochemistry and Health 26:89–95.

Zagal, E., C. Muñoz, M. Quiroz, and C. Córdova. 2009. Sensitivity of early indicators for evaluating quality changes in soil organic matter. Geoderma 151:191–198. doi:10.1016/j.geoderma.2009.04.004

Zibilske, L. 1994. Carbon mineralization. In: S.H. Mickelson, editor, Methods soil analysis: Part 2. Microbiological biochemical properties. SSSA, Madison, WI. p. 835–863.